SOFTWARE TESTING USING THE CODE REVIEW TECHNIQUE: AN EXPLORATORY STUDY

Abstract

This paper presents an exploratory study with Belbin roles, in particular role types, in an individual activity consisting of code review. The objective is to identify if, in addition to any of the three types of roles, the position of the fault and the gender of the subject influence the activity of detecting faults in the code. To create an experimental context for the review process, during a work session the subjects, who were software engineering students, used a code with injected faults for their review. With respect to the types of roles, the results of the experiment do not show significant differences either in the efficiency index obtained by the subjects in the testing process, or with the confusion index of failures. On the other hand, regarding the position of the fault, the results show significant differences between the faults detected in the first half of the code with respect to the remaining second half. Regarding the gender of the subject, the experiment does not show a significant difference between the detected errors. At the end of this report, researchers should perform a short-term analysis of the errors introduced in the code to obtain a better version of the experimental object, allowing them to perform a second controlled experiment under less restrictive conditions.